Early event in maize leaf epidermis formation as revealed by cell lineage studies

The epidermis of the grass leaf is a single cell layer organized as longitudinally oriented parallel rows of cells that are elongated in the direction base to leaf tip. Starting from the primordium, the cells of the epidermal layer divide according to planes that allow the epidermis to grow in the transverse direction, via longitudinal anticlinal divisions, or in the longitudinal direction, via transverse anticlinal divisions. Both anatomical studies (Esau, 1977) and the type of leaf sectoring reported in grass species (Tilney-Basset, 1986; Klekowski, 1988) support the view that longitudinal anticlinal divisions are mainly restricted to early stages of leaf development.

The organization of cells in longitudinal rows is typical also for the internal layers of the leaf. In maize, for example, the leaf is divided into longitudinal units by parallel veins located in the mesophyll layer (Sharman, 1942; Esau, 1943; Russell and Evert, 1985; reviewed by Langdale et al., 1989 and Freeling, 1992). Veins have been classified as mid, lateral, intermediate and small (Sharman, 1942). They derive from the central layer of the leaf primordium (Langdale et al., 1989). The adaxial and the abaxial epidermis and the middle mesophyll layer show coordinate development (Freeling, 1992; Freeling and Lane, 1993): cells with a particular shape are present on the epidermis in positions corresponding to the location of veins. In short, when maize leaves are viewed from above, vein positions are marked on the epidermis by costal long cells (Freeling and Lane, 1993), which constitute the structures known as ribs.

The epidermis of the young maize shoot is covered by the juvenile wax layer (Salamini, 1963). Mutations mapping to at least 8 different genetic loci modify the shape and drastically reduce the wax layer (summarized by Bianchi et al., 1985). Wax extrusion is cell autonomous, as is evident from the variegation pattern of leaves of somatically unstable mutants (Maddaloni et al., 1990). Mutable alleles of the glossy-1 locus revert both somatically and germinally to the wild-type phenotype (Maddaloni et al., 1990; Bossinger et al., 1992).

We have used the mutant glossy-1 mutable 8 (gl1-m8) to study the width and distribution of revertant sectors. The leaf epidermis is quite suited to such analyses because it originates from a single meristematic layer (Sharman, 1942), and because the start and end of revertant sectors can be defined with respect to ‘landmark’ signals represented by vein-rib boundaries. When using transposon-induced sectors in cell lineage analysis, a requisite is that the excision of the transposon is not developmentally regulated. To avoid this, only sectors that originate in the apical meristem before leaf promordia are formed should be studied. The inception of leaf primordia, in fact, induces differentiation between cells. Steffensen (1968) has shown that sectors with a size from 5 to 34% of the midrib to margin space − fully comparable with those studied by us − always appear in more than one leaf, and are assumed to have been generated by cells present on the shoulder of the meristem before leaf primordia inception.

Additional proof that in our system excisions are random in position is given by Maddaloni et al. (1990), who noted that single cell sectors were spread randomly over the epidermal surface, while sectors as large as those studied in this paper appeared in more than one leaf, as was the case for the sectors described by Steffensen (1968).

Three developmental problems are addressed. The first concerns the existence of leaf epidermal compartments. The concept of compartment (Garcia-Bellido and Merriam, 1971; Garcia-Bellido et al., 1973; 1976) refers to the observation that cellular clones do not cross a line that defines the border between morphologically distinct domains. It is accepted that epidermal segments of Drosophila are subdivided into com-partments, developmental units expressing a specific set of homeotic genes (Brower, 1985; Lawrence and Morata, 1993). In our system candidates for compartment boundaries are the epidermal ribs.

The second question addresses the possibility of recognizing the number of cells that have a founder role during the early development of the leaf epidermis. This role can be clarified because of the particular position these cells occupy with respect to the ribs in the primordia or in adult leaves; as founders they generate groups of cells that are clonally related.

Thirdly, we have attempted to gain information on the polarity of longitudinal anticlinal cell divisions needed to add, in the transverse direction of the leaf, cells to leaf width. Starting with one cell the problem is to establish which out of several models of cell division (see Fig. 1), fits best the distribution and width of revertant sectors observed.

The maize mutant gl1-m8 was isolated in an attempt to tag the Glossy-1 locus by transposon mutagenesis (Maddaloni et al., 1990). The stock waxy mutable 7 (wx-m7), where an active copy of the activator (Ac) transposon (McClintock, 1951) is inserted into the Wx gene, was the transposon donor. The allele gl1-m8, however, was not generated by the insertion of Ac but, nevertheless, it behaved autonomously, i.e. another self-excising element was present at the locus (Maddaloni et al., 1990). The gl1-m8 mutant is characterized by its somatic instability: reversions to the wild-type phenotype (longitudinal ‘sectors’) are present on both the leaf sheath and blade (Bossinger et al., 1992).

Four hundred seedlings of the gl1-m8 strain were grown at 25°C under natural light conditions supplemented with 16 hours per day of artificial illumination (1900 μE m⁻² second⁻¹). Revertant sectors were studied in 344 leaves, of which 88, 156 and 100 were, respectively, on the first, second and third leaves of the young shoot. Sectors were localized on both the adaxial and abaxial leaf surfaces; when, as usual, both surfaces of a leaf were sectored in corresponding positions, only the adaxial sector was studied.

Leaves were dissected from the shoot as soon as the tip of the fourth leaf appeared. Each leaf was inspected under water to determine whether revertant sectors extended along the whole blade length (Bossinger et al., 1992). A 1 cm strip was cut transversely where the leaf blade width was at its maximum. Samples were mounted on a stub, coated with gold and studied with a Hitachi S-2300 scanning electron microscope (SEM). Under these conditions, the epidermal cells tended to collapse, with the exception of those constituting the mid, lateral and intermediate ribs. Revertant epidermal sectors covered by the wax layer were easily recognized and classified. Where necessary, more precise details of the leaf surface (see Fig. 3) were obtained from leaf samples infiltrated with 7.4% formaldehyde, 5% acetic acid and 50% ethanol, fixed overnight at 4°C, dehydrated with dimethoxymethane (DMM) for 24 hours, dried in liquid carbon dioxide, mounted on a stub and coated with gold.

The maize leaf blade possesses mid, lateral and intermediate ribs. The midrib consists of 8-12 rows of cells more elongated in the direction base to leaf tip than the adjacent leaf blade cells. These are the costal long cells. Between the midrib and each leaf margin, 5 lateral ribs are present (Fig. 2A). These lateral ribs consist of two rows of costal long cells. The space between two consecutive lateral ribs is defined as a leaf developmental module. The module (Fig. 2B) is divided into parts α and β by the intermediate rib, represented by a single row of costal long cells.

In young leaf blades, 12 modules make up the epidermis. The nomenclature concerning leaf modules is given in Fig. 2A and a module is detailed in Fig. 2B. Leaf modules with revertant sectors on the adaxial or abaxial leaf surface were studied according to Fig. 2B: leaves sectored on the adaxial side, right from the midrib (R), or on the abaxial side, left from the midrib (L), were positioned with their tips pointing upwards; leaves sectored in L on the adaxial side or in R on the abaxial side were positioned with the tip downwards.

The leaf developmental module has a mean width of 53 cells. Cell numbering was assingned starting in the half module α with 1α for the first cell following the lateral rib; the other 25 cells were given the numbers 2α to 26α. The 27 cells of β were numbered from 1β (the first cell after the intermediate rib) to 27β. The two costal long cells of the lateral rib distal to the midrib were marked 26β and 27β; the intermediate costal long cell was given the number 26α.

Epidermal sectors are clones of cells covered by the juvenile wax layer. Two types of sectors (A and B) were studied: sectors of type A extended transversally for ⁠, 1 or more modules; sectors of type B were smaller than a module. Because sectors not starting at the precise borders of the module were commonly observed, and because between modules differences in width were noted, a sector was defined as type A when its minimum width was of 22 cells.

The data recorded for the A and B sectors were, beside width, the cell number at the start and end, and the position of the module(s) on the leaf surface (right (R) or left (L) with respect to the midrib; Fig. 2A). In assigning cell position of the start, counting took place from the left in both half-modules; the cell number at the end of sectors was assigned in a similar way, with the exceptions of sectors ending at 22α to 26α or at 23β to 27β: for these, cell numbering was from the right.

The ribs divide the juvenile leaf blade into repetitive modules. The module contains the cells betwen two lateral ribs, the lateral rib distal with respect to the midrib, and the intermediate rib cell (Fig. 3A). Intermediate and lateral ribs consist of rows of 1 and 2 costal long cells, respectively (Fig. 3B, C). In lateral ribs nearer to the midrib, the number of rows of long costal cells is, in rare cases, larger than two.

Six developmental modules are present on both the right and the left side of the leaf blade (Fig. 2A). The modules near the leaf borders are not flanked by lateral ribs. Hairs mark the cell files at the leaf border; these hairs are oriented in the direction of the tip of the leaf blade, a useful feature in microscopic analysis. The module width (number of cells in the transverse direction) was counted for each of the 12 units present on the abaxial and adaxial surfaces of the first three leaves. Because adaxial and abaxial module width were not significantly different, only the mean width of modules is reported (Table 1). Module width varied between 40 and 53 cells. RA and LA (see Fig. 2) were exceptional because of their reduced width of about 20 cells. In these two modules, in a few cases, the intermediate rib was not precisely positioned in the centre. The presence of intermediate ribs with more than one long costal cell, or of more than one intermediate rib, was also noted in RA and LA. Sectors present on such exceptional modules were not considered. Compared with other modules, the modules at the border of the leaf had a slightly reduced and more variable width. The width of module LC, third leaf (53 cells), was taken as standard.

A sector type A is a somatic clone derived from a cell that experienced a reversion event from gl1-m8 to Gl1, and that occupies at least half a module (see also footnotes of Table 2). B sectors occupy less than half a module.

In total, 292 A sectors were considered and Fig. 4 shows the types most commonly found. A sector covering precisely one module (from 1α to 27β) is shown in Fig. 4A. Sectors precisely limited to the half-modules α or β were also noted (Fig. 4B). Fig. 4C shows a sector that started at 1α but ended within β. An A sector crossing a lateral rib is illustrated in Fig. 4D.

A classification of A sectors is presented in Table 2. Preliminary analyses of the width and distribution of A sectors revealed only marginal differences among leaves or between their abaxial and adaxial surfaces (results not shown). Sectors of type A were progressively larger when starting nearer to the midrib. Sectors with a width corresponding to 1 or modules were frequent; also frequent were those with a width between and 1,1 and and and 2 modules. It was note-worthy that out of the 276 A sectors with a width from to 4 modules, 100 had a width corresponding to a module or to multiples thereof (see, however, footnote 1 of Table 2).

One hundred and twenty two type B sectors were scored. Representative types are reproduced in Fig. 4, covering half of α (Fig. 4E), or crossing the intermediate rib (Fig. 4F), or ending precisely at the intermediate or lateral ribs (Fig. 4G,H).

In the last two columns of Table 2, A sectors are classified based on their origin, either in α or in β. As a rule, type A sectors showed a strong tendency to start within the module. Exceptions were the class and, in part, the 1 class. The precise position of start and end of the 292 A sectors within the developmental module is summarized in Fig. 5. The figure consists of a reproduction of the standard module (with the midrib at the left), and of histograms representing the number of sectors starting (upper) or ending (lower) at a given cell number.

The preferred start position of A sectors was cell 1α (90 cases out of 292). Other preferential start cells were 26β, 27β, 2α, 26α and 1β (respectively, 13, 11, 9, 9 and 10 sectors). These are all cell positions flanking cell 1α, or correspond to the lateral or intermediate rib cells, or are located near the initiation of the half-module β. Sectors type A most frequently ended at cells 25β, 26β and 24α, 25α, 26α. Among the 74 sectors whose ends were positioned at 27β, 56 were those that extended up to the leaf border. The number of A sectors starting around 1β was significantly lower than the number of those ending in the proximity of this cell; in the module, this cell follows the costal long cell of the intermediate rib.

Fig. 6 presents start and end points for the 122 B sectors. Start points were most frequently around positions 1α and 1β. Moreover, in α, start points between 2α and 26α were less numerous than in β at the homologous positions 2β to 27β. B sectors ended preferentially before cell 1α and 1β, with minor peaks around cells 9α, 12α, 2β, 9β, and 22β.

The data available for B sectors were used to study the relationship between the width and position of the start of sectors (Fig. 7A). In α, a gradual decrease of width was evident when the starting point of sectors moved from 1α to 26α. This was less evident in β. The width of B sectors was also plotted against their observed number (Fig. 7B). If the exponential type of cell division shown in Fig. 1 were to operate during epidermis formation, the preferred sector widths should be of 1, 2, 4, 8, 16 cells. The data in Fig. 7B contradict this interpretation.

Sussex (1989) and Walbot (1985) have summarized the relevant developmental differences that exist between plants and animals. Among these, the lack of a sequestered germline, the continuous embryogenic state of meristems and the developmentally late commitment of cells to a specific fate, are characteristics of plant development. However, perhaps the unique cellular feature, which has the most profound implications on plant cell growth and differentiation, is that they are encapsulated by a rigid cell wall. This particular condition prevents cell rotation and migration and increases the morphogenetic importance of early cell divisions which already contribute to defining adult cell arrangements. Our results define the role played by early longitudinal anticlinal divisions in shaping the modular nature of leaf lamina epidermis.

In the introduction to this paper we have stressed that a very large proportion of the sectors considered originate from cells present in the meristem before leaf primordia are initiated. This allows us to conclude that early transposon excision is apparently not influenced by a specific state of cell differentiation. The data provided, which demonstrate that sectors preferentially start and end at cellular positions corresponding to the borders of developmental modules, support the conclusion that cells, which by longitudinal anticlinal divisions originate a clone of reverted tissue, have a defined role in the formation of the developing module already at an early stage of primordium differentiation. These cells are considered ‘module founder cells’ and alternative possibilities for their early positioning in the module are described in Fig. 8A. The start positions of sectors, together with sector widths (Fig. 8B), can be explained by the existence of at least 4 founder cells, all derived from one progenitor. With one exception, these cells should arise from divisions adding cells to the width of the leaf in the direction from the midrib towards the leaf margin. Sectors of type A larger than module have, in fact, a strong tendency to start in the midrib proximal half-module α (Table 2): β seems, in a sense, to be hierarchically dependent on α as the site for the origin of sectors. This suggests a clonal derivation of the cell(s) founding β from the one(s) starting the module in α, supporting a polarity of development in the direction α to β.

A partial exception to the rule is the behaviour of the -1 A sectors.

Our findings are consistent with the notion that leaf blade expansion proceeds from midrib to margin (Sharman, 1942; Esau, 1943; Freeling, 1992). Becraft and Freeling (1991) demonstrate that putative morphogens involved in blade differentiation move in the direction midrib to margin. The inception of leaf primordia in the apical meristem starts also where the future midrib will differentiate (position marked by the presence of a major procambial strand) and proceeds around the meristem dome in two, opposite, directions (Sharman, 1942); the same applies to ligule differentiation, which proceeds from the midrib toward the margins (Hake et al., 1985).

Data concerning B sectors help to elucidate the types of longitudinal anticlinal cell divisions that are taking place within half-modules α and β. The available data indicate that in the α-sector, the sector width decreases when the starting point moves from 1α to 26α. In β this relationship holds true particularly for the right part of the half-module. Such behaviour is compatible with both the polarized and the stem-cell-like type II models of cell division shown in Fig. 1. Experiments based on the use of two epidermal markers may indicate, in the future, which model is more correct. The stem-cell-like type I and the exponential models are excluded, the first because it predicts a similar width for all sectors, and the second because the sectors should preferentially have a width of 1, 2, 4, 8, and 16 cells, a prediction contradicted by the data of Fig. 7B.

In this paper we provide results in favour of a clonal type of development during early leaf epidermis formation. The sectors studied extend longitudinally along the whole leaf blade and reflect events occurring early in leaf primordia development, often prior to the formation of epidermal ribs (Langdale et al., 1989; Poethig, 1984). Because type A sectors frequently occupy precisely half of the entire module, this allows the ribs to be considered as compartment boundaries. The clonal relationship between α and β half modules restricts the consideration of compartment boundaries to lateral ribs. These boundaries are frequently respected by epidermal clones; for instance, the location where the proximal border of a sector is positioned has a high probability of being the first cell after a lateral rib. In sectors covering more than 1 module, however, lateral ribs do not represent effective boundaries. This can be explained by proposing that large epidermal sectors reflect a mutated state of a large fraction of the apical meristem (Bossinger et al., 1992).

The identification of the vein-rib as a line separating clonally unrelated cells is not new. Sectored maize seedlings with bilateral symmetry, where the boundary between the contrasting phenotypes coincides with the leaf midrib, have been described (Steffensen, 1962; Coe and Neuffer, 1978; Bossinger et al., 1992). Christianson (1986) has also provided data in favour of the existence of compartments in cotton cotyledons. In his experiment the boundaries between two tissues different in genotype corresponded to veins, or were halfway between. Although in 50% of cases the sectors failed to exactly fill a putative clonal compartment − because of either a deficiency or an excess of small fan-shaped subclones − Christianson’s results can be accepted as proof for the existence of compartments.

In Drosophila, the final proof of the existence of clonal compartments was obtained by using mutations that gave a growth advantage to somatic sectors (Garcia-Bellido et al., 1976). For the dorsal-ventral boundary of the wing margin of Drosophila, exceptions have, however, been described where the compartment boundaries are crossed by cellular clones (Garcia-Bellido et al., 1976; Morata and Lawrence, 1979). The case of maize leaf epidermis discussed here shows more flexible relationships between boundaries and cellular clones. A particular case of flexibility is the positioning of sector borders around cells 1α and 1β, a positioning ‘almost’ precise but not restricted to the two cells mentioned. Such cases suggest that lateral ribs may differentiate after the maturation of veins has determined the positioning in the meristem of the module founder cells. Alternatively, the ribs should be considered as the end product of a differentiation process stimulated by the vein and capable by itself of preventing further cell divisions. In the maize leaf, the position in the mesophyll of veins seems, in any case, to have a critical morphogenetic role in the development of the epidermis. This observation favours the possibility that veins already have their initials topographically imprinted in the meristem at the time when leaf primordia are in the process of being formed (discussed by Freeling et al., 1988).